Improving Search in Tele-Lecturing: Using Folksonomies as Trigger to - - PowerPoint PPT Presentation

improving search in tele lecturing using folksonomies as
SMART_READER_LITE
LIVE PREVIEW

Improving Search in Tele-Lecturing: Using Folksonomies as Trigger to - - PowerPoint PPT Presentation

Improving Search in Tele-Lecturing: Using Folksonomies as Trigger to Query Semantic Datasets to Extract Additional Metadata International Conference on Web Intelligence, Mining and Semantics (WIMS 2011) Franka Moritz, Maria Siebert and


slide-1
SLIDE 1

Improving Search in Tele-Lecturing: Using Folksonomies as Trigger to Query Semantic Datasets to Extract Additional Metadata

International Conference on Web Intelligence, Mining and Semantics (WIMS 2011)

Franka Moritz, Maria Siebert and Christoph Meinel Hasso-Plattner-Institute University of Potsdam

slide-2
SLIDE 2

Improving Search in Tele-Lecturing | Moritz, Siebert, Meinel | WIMS 2011

Agenda

■ Description of the tele-TASK Project ■ Motivation ■ Possibilities for embedding semantic information ■ Examples of the usage of the data □ Search function □ Recommandation function ■ Conclusion

slide-3
SLIDE 3

Improving Search in Tele-Lecturing | Moritz, Siebert, Meinel | WIMS 2011

Research project: tele-TASK

■ System for recording lectures ■ Records two video and one audio stream ■ Used since 2001 ■ More than 3200 lectures recorded ■ More than 8000 video podcasts provided

slide-4
SLIDE 4

Improving Search in Tele-Lecturing | Moritz, Siebert, Meinel | WIMS 2011

tele-TASK Web Portal http://www.tele-task.de

slide-5
SLIDE 5

Improving Search in Tele-Lecturing | Moritz, Siebert, Meinel | WIMS 2011

Motivation

slide-6
SLIDE 6

Improving Search in Tele-Lecturing | Moritz, Siebert, Meinel | WIMS 2011

Embedding DBPedia into the tele-TASK portal meta data

Different possibilities for data sources

slide-7
SLIDE 7

Improving Search in Tele-Lecturing | Moritz, Siebert, Meinel | WIMS 2011

Usage of the DBpedia data

■ Extraction of keywords ■ Search for representation in DBpedia ■ Parsing information from DBpedia ■ Adding information to the database □ Synonyms or generalizations of terms □ Desciptions of a term ■ Using the parsed information in different projects

slide-8
SLIDE 8

Improving Search in Tele-Lecturing | Moritz, Siebert, Meinel | WIMS 2011

Definition of Tags

slide-9
SLIDE 9

Improving Search in Tele-Lecturing | Moritz, Siebert, Meinel | WIMS 2011

Extending the Search function

■ Adding additional data to searchable fields □ Generalizations of tags □ Synonyms of tags ■ Search for tags with the search term in the description ■ Extending the search request □ e.g. using synonyms

slide-10
SLIDE 10

Improving Search in Tele-Lecturing | Moritz, Siebert, Meinel | WIMS 2011

Extending the Recommendation Function – Synonyms (1)

Synonyms: ■ Synonyms create an equivalence relation ■ Each group of synonyms generates an equivalence class ■ Interpret tags in one equivalence class as one word

slide-11
SLIDE 11

Improving Search in Tele-Lecturing | Moritz, Siebert, Meinel | WIMS 2011

Extension of the Recommendation function - Synonyms (2)

■ The normal similarity function for tags is extended to a more complex version ■ Each equivalence class of tags is treated like one tag in the simple calculation function

slide-12
SLIDE 12

Improving Search in Tele-Lecturing | Moritz, Siebert, Meinel | WIMS 2011

Extension of the Recommendation function - Generalization (1)

■ Generalization creates partial

  • rder

■ Distance between objects is unknown ■ Defining sets of parents and children ■ Calculating the distance of two

  • bjects in the tree as dist(s,t)

Animal Bird Insect Finch Stork Bee Mosquito Chaffinch

slide-13
SLIDE 13

Improving Search in Tele-Lecturing | Moritz, Siebert, Meinel | WIMS 2011

Extension of the Recommendation function - Generalization (3)

■ Using distance as weigth for calculation of similarity ■ Using the weigths of the tags

slide-14
SLIDE 14

Improving Search in Tele-Lecturing | Moritz, Siebert, Meinel | WIMS 2011

Conclusion and Future Work

■ Implementation for tags was shown ■ More triggers have to be included ■ Problems to solve: □ To few tags to generate for a good result evaluation ◊ Advancement of user interest ◊ Easier access to the functions □ The results are not visible to the user

slide-15
SLIDE 15

|

Thank you for your attention.

Maria Siebert Hasso-Plattner-Institute University of Potsdam Potsdam, Germany